Diamond nv color center vector magnetic field measurement method, system, device and medium based on spatiotemporal tensorization denoising and feature extraction

Abstract

The present application relates to the technical fields of quantum precision measurement and high-precision signal processing, and discloses a diamond NV color center vector magnetic field measurement method, system, device and medium based on space-time tensorization noise reduction and feature extraction, which comprises the following steps: constructing a three-dimensional data tensor, performing dynamic self-adaptive threshold binaryzation processing on the three-dimensional data tensor to generate a binaryzation logic tensor; performing three-dimensional XOR logic filtering and Hamilton eigenvalue trajectory checking on the binaryzation logic tensor to obtain a target tensor after noise reduction; extracting the resonance frequency values corresponding to each crystal axis from the target tensor, and jointly solving the vector magnetic field and temperature through an overdetermined nonlinear equation set. The present application can construct a three-dimensional space-time tensor by using an FPGA under a continuous wave magnetic measurement mode, realize efficient parallel extraction of eight resonance peaks, and greatly improve the vector magnetic field measurement precision and the robustness of the system under a complex environment.

Description

Technical Field

[0001] This invention relates to the fields of quantum precision measurement and high-precision signal processing technology, and in particular to a method, system, device and medium for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction. Background Technology

[0002] With the rapid development of quantum sensing technology, magnetometers based on diamond nitrogen-vacancy (NV) centers have shown great application potential in fields such as fundamental physics, biomedical imaging, power system current monitoring, and geomagnetic navigation due to their high sensitivity, high spatial resolution, and good biocompatibility at room temperature. The energy levels of NV centers undergo Zeeman splitting under the influence of an external magnetic field. By using optically detected magnetic resonance (ODMR) technology, sweeping microwave frequencies, and detecting the depression points (resonant frequencies) of the fluorescence signal, information about the external magnetic field can be retrieved.

[0003] In related technologies, continuous wave magnetic field measurements of diamond NV centers typically rely on scalar magnetic field measurement methods. For example, one existing system for measuring the magnetic field of diamond NV centers based on data matrix analysis converts swept fluorescence signals into a two-dimensional matrix and extracts the effective signal using block matrix operations and XOR logic within an FPGA. This method effectively improves the measurement speed of uniaxial (scalar) magnetic fields and reduces hardware overhead.

[0004] However, in real-world complex industrial environments (such as high-voltage power grid environments and mobile platform environments), magnetic fields often exhibit three-dimensional spatial vector characteristics, accompanied by severe temperature drift and broadband electromagnetic noise. Diamond has four equivalent NV-axis orientations, which, under a three-dimensional vector magnetic field, will split into eight different ODMR resonant frequency peaks. Traditional two-dimensional data matrices or simple peak search algorithms have the following serious shortcomings when processing such complex signals: First, there is the issue of overlapping peaks and spurious peak interference. When the direction of the external vector magnetic field changes, the resonant peaks of different crystal axes may approach or overlap each other. Conventional threshold judgment and two-dimensional matrix XOR logic can easily misjudge noise spikes as effective resonant frequencies.

[0005] Second, there is a lack of joint analysis of temperature and multi-axis correlation. Two-dimensional analysis often only focuses on frequency sweep data of a single time period, ignoring the spatiotemporal correlation between multiple channels and multiple periods, and cannot simultaneously calculate the accurate three-dimensional components of the vector magnetic field and temperature parameters under strong background noise.

[0006] Third, the dynamic tracking capability is weak. Due to the constant fluctuations in the noise floor and baseline under dynamic environments, the existing noise reduction circuit and differentiator combination is prone to distortion of the output modulation signal when facing non-Gaussian dynamic noise, which leads to the failure of subsequent FPGA matrix processing. Summary of the Invention

[0007] In view of the aforementioned existing problems, the present invention provides a method, system, device and medium for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction.

[0008] Therefore, the technical problem solved by the present invention is that the existing technology has shortcomings in vector magnetic field measurement, such as poor noise resistance, easy misjudgment of multi-peak overlap, and inability to simultaneously achieve high-precision composite measurement of vector and temperature.

[0009] To solve the above technical problems, the present invention provides the following technical solution: a method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction, comprising: constructing a three-dimensional data tensor, the dimensions of which correspond to the time period dimension, the frequency sweep point dimension, and the NV color center crystal axis projection space dimension, respectively; The three-dimensional data tensor is subjected to dynamic adaptive threshold binarization to generate a binary logical tensor; the dynamic adaptive threshold is determined based on the local dynamic baseline mean and the adaptive error threshold under a set confidence level. Perform three-dimensional XOR logic filtering and Hamiltonian eigenvalue trajectory verification on the binarized logic tensor to obtain the denoised target tensor. The resonant frequency values ​​corresponding to each crystal axis are extracted from the target tensor, and the vector magnetic field and temperature are jointly solved by the overdetermined nonlinear equations.

[0010] As a preferred embodiment of the diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction described in this invention, the dynamic adaptive threshold binarization process includes dynamic estimation of the local dynamic baseline mean and the adaptive error threshold by a combination of a local frequency sliding window and a time history buffer window. The local dynamic baseline mean is calculated using a robust statistical method based on the median and absolute median difference, and the adaptive error threshold is generated by multiplying the local noise scale obtained based on the robust statistics with a statistical threshold coefficient determined according to a set confidence level.

[0011] As a preferred embodiment of the diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction described in this invention, wherein: the frequency local sliding window covers a preset neighboring frequency point with the current frequency point as the center, and the time history cache window includes the current scanning cycle and a preset number of previous scanning cycles; The mean of the local dynamic baseline is the median of the data within the local frequency sliding window, and the local noise scale is obtained by multiplying the median of the absolute deviation of the data within the local frequency sliding window from the median by a fixed constant factor.

[0012] As a preferred embodiment of the diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction described in this invention, the execution of three-dimensional XOR logic filtering and Hamiltonian eigenvalue trajectory verification includes performing XOR logic operation on the binarized logic tensor in adjacent scanning cycles along the time dimension, and determining the elements with an XOR result of 1 as random burst noise and setting them to zero. Along the spatial dimension, Hamiltonian eigenvalue trajectories are verified for the retained candidate points. Candidate points whose residuals with the predicted frequency trajectory exceed a preset threshold are judged as deviating from physical constraints and set to zero.

[0013] As a preferred embodiment of the diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction described in this invention, the Hamiltonian eigenvalue trajectory verification includes temporal continuity prediction and spatial geometric consistency constraint; wherein, the temporal continuity prediction predicts the frequency of the current candidate point based on the frequency extracted in the previous cycle and the frequency drift velocity, and the spatial geometric consistency constraint is based on the eigenvalue trajectory determined by the projection relationship between the four NV crystal axis directions and the applied vector magnetic field.

[0014] As a preferred embodiment of the diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction described in this invention, the method of jointly solving the vector magnetic field and temperature through an overdetermined nonlinear equation system includes: using half of the difference between the positive and negative resonant frequencies corresponding to the four crystal axes as the differential modulus, constructing a matrix equation proportional to the projection of the vector magnetic field, and solving for the external vector magnetic field components using a least-squares matrix pseudo-inverse algorithm; simultaneously, using half of the sum of the positive and negative resonant frequencies as the common modulus, separating the change in the zero-field splitting parameter, and then solving for the temperature change.

[0015] As a preferred embodiment of the diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction described in this invention, wherein: the least squares matrix pseudo-inverse algorithm is calculated based on a geometric matrix composed of unit vectors of the four NV crystal axis directions; the temperature change is obtained by dividing the change of the zero-field splitting parameter by the temperature coefficient.

[0016] This invention provides a diamond NV color center vector magnetic field measurement system based on spatiotemporal tensor denoising and feature extraction.

[0017] As a preferred embodiment of the diamond NV color center vector magnetic field measurement system based on spatiotemporal tensor denoising and feature extraction described in this invention, it includes: an optical path excitation module, a vector microwave control module, a three-dimensional probe module, and a tensor magnetometry electronics module; The optical path excitation module has an acousto-optic modulator, a first lens, a dichroic mirror, and an objective lens arranged sequentially in front of the laser source; the laser emitted by the laser source is modulated and reflected, and then focused by the objective lens onto the diamond NV color center sample in the three-dimensional probe module. The three-dimensional probe module includes: a three-dimensional orthogonal Helmholtz bias coil, a microwave radiating antenna, a long-pass filter, and a diamond NV color center sample ensemble placed at the center; the three-dimensional orthogonal Helmholtz bias coil is used to apply a known three-dimensional bias static magnetic field to separate the degenerate energy levels of the four crystal axes of the NV color center; The vector microwave control module includes: a multi-channel microwave source, a radio frequency switch, a broadband power amplifier, and a system control PC; the multi-channel microwave source generates a swept-frequency microwave signal covering different crystal axis resonant ranges, which is then fed into the microwave radiating antenna after being amplified by the power amplifier; The tensor magnetometry electronics module includes: a photodetector (PD), a transimpedance amplifier, a programmable bandpass filter, a multi-channel analog-to-digital converter (ADC), an FPGA tensor processing module, and a digital-to-analog converter (DAC). The red fluorescence signal emitted by the diamond NV color center sample passes through the dichroic mirror and long-pass filter, is received by the photodetector and converted into an electrical signal, and is then pre-conditioned by the transimpedance amplifier and the programmable bandpass filter before being input to the multi-channel analog-to-digital converter module. The multi-channel analog-to-digital converter module converts the conditioned continuous fluorescence voltage signal into a discrete digital signal and sends it to the FPGA tensor processing module. The system control PC synchronously sends the three-dimensional bias static magnetic field parameters, microwave sweep period, and frequency point data to the FPGA tensor processing module.

[0018] The present invention provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of a method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction.

[0019] The present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of a method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction are implemented.

[0020] The beneficial effects of this invention are as follows: This invention breaks through the limitations of traditional single-period or two-dimensional matrices, constructing a three-dimensional spatiotemporal tensor (time × frequency × crystal axis space). By utilizing the XOR logic operation of adjacent time dimensions of the tensor, random burst impulse noise in industrial and natural environments is filtered out extremely effectively, greatly reducing the false peak misjudgment rate.

[0021] When the vector magnetic field changes cause the ODMR resonance peaks of different crystal axes to approach or overlap each other, this invention uses tensor constraint verification of spatial dimension and relies on the geometric correlation of the four crystal axes for filtering, so that the true frequency can still be extracted with high accuracy in complex cross-overlapping spectral lines.

[0022] Traditional two-dimensional matrix methods are generally used to extract differential-mode signals to calculate uniaxial magnetic fields. This invention, through the full tensor solution of overdetermined equations at eight resonant frequencies, can losslessly separate the common-mode variation components, thereby accurately retrieving the real-time changes in ambient temperature while calculating the three-dimensional vector magnetic field strength, achieving precise composite measurement of magnetic field and temperature. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of a method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction, provided as an embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram of the overall structure of a diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction, provided in an embodiment of the present invention.

[0026] Figure 3 The experimental data verification diagram shows the comparison between the method of the present invention, which is based on spatiotemporal tensor denoising and feature extraction for measuring the magnetic field of diamond NV color centers, and the traditional matrix method for extracting resonance peaks under strong noise, in an embodiment of the present invention. Detailed Implementation

[0027] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0028] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction, including: S1: Construct a three-dimensional data tensor, with dimensions corresponding to the time period dimension, the frequency sweep point dimension, and the NV color center crystal axis projection space dimension, respectively.

[0029] S2: Perform dynamic adaptive threshold binarization on the three-dimensional data tensor to generate a binary logical tensor; the dynamic adaptive threshold is determined based on the local dynamic baseline mean and the adaptive error threshold under a set confidence level.

[0030] S3: Perform three-dimensional XOR logic filtering and Hamiltonian eigenvalue trajectory verification on the binarized logic tensor to obtain the denoised target tensor.

[0031] S4: Extract the resonant frequency values ​​corresponding to each crystal axis from the target tensor, and solve the vector magnetic field and temperature together through the overdetermined nonlinear equations.

[0032] It should be noted that by constructing a three-dimensional data tensor that integrates the time period, frequency sweep points, and crystal axis spatial dimensions, and employing robust dynamic baseline estimation with a frequency-time joint window and adaptive confidence threshold binarization, the influence of non-stationary noise such as laser drift and broadband electromagnetic interference on the ODMR signal is effectively suppressed, significantly enhancing the anti-interference capability in complex environments. Furthermore, three-dimensional XOR logic filtering is used to eliminate random burst noise along the time dimension, and Hamiltonian eigenvalue trajectory verification is introduced to eliminate false peaks deviating from physical constraints along the spatial dimension, solving the misjudgment problem caused by multi-peak overlap and achieving high-accuracy parallel extraction of eight resonance peaks. Finally, by constructing a differential-mode and common-mode overdetermined equation set and employing least-squares pseudo-inverse solution, high-precision vector magnetic field components and temperature changes can be simultaneously separated under continuous-wave magnetic measurement, achieving high-precision composite measurement of vector magnetic field and temperature.

[0033] Example 2 is an embodiment of the present invention. Based on the above embodiment, a method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction is provided.

[0034] Furthermore, in this embodiment, step S1 constructs a three-dimensional data tensor, with dimensions corresponding to the time period dimension, the frequency sweep point dimension, and the NV color center crystal axis projection space dimension, respectively. Specific steps include: In continuous Within each frequency sweep cycle, for the frequency sweep intervals corresponding to the four NV crystal axes, the FPGA spatiotemporal tensor construction unit receives the fluorescent discrete digital signals. and the corresponding microwave frequency Integrate and construct a 3D data tensor , among which dimension Indicates the time period dimension. This indicates the frequency point dimension of a single frequency sweep. Represents the dimension of the crystal axis projection space ( This corresponds to the projection separation channels of the four crystal axes. Through this tensor quantization, the signal acquires historical memory properties and spatial correlation properties.

[0035] Furthermore, in this embodiment, step S2 performs dynamic adaptive threshold binarization on the three-dimensional data tensor to generate a binarized logical tensor; the dynamic adaptive threshold is determined based on the local dynamic baseline mean and the adaptive error threshold under a set confidence level, and the specific steps include S201-S204: S201: The local dynamic baseline mean and adaptive error threshold are dynamically estimated by a combination of a local frequency sliding window and a time history buffer window; For each slice in tensor X, calculate the local dynamic baseline mean and variance for the current period, and generate an adaptive error threshold Δ(m,k) based on the set confidence level. If the fluorescence intensity of tensor element X(m,n,k) is within ±Δ(m,k) of the baseline minimum extreme point, then assign the corresponding logical value tensor β(m,n,k) to 1; otherwise, assign it to 0, thus obtaining the binarized logical tensor β.

[0036] S202: The local dynamic baseline mean is calculated using a robust statistical method based on the median and absolute median difference. The adaptive error threshold is generated by multiplying the local noise scale obtained based on this robust statistics with the statistical threshold coefficient determined according to the set confidence level.

[0037] The frequency local sliding window covers a preset neighboring frequency range centered on the current frequency point, while the time history buffer window includes the current scan cycle and a preset number of previous scan cycles; for the first... The scan cycle, the first The first crystal axis channel Each frequency point defines a local frequency window. Wherein, the window length is , where r is the window radius; in FPGA implementation, a value of 9 to 33 frequency points is preferred. This window is used to estimate the local fluorescence background and noise levels near the current frequency point.

[0038] Simultaneously define a time history window: in, This is the length of the historical cycle buffer; a preferred value is 4 to 16 scan cycles. The time window is used to enhance baseline continuity and robustness in dynamic environments.

[0039] S203: The local dynamic baseline mean is the median of the data within the local frequency sliding window. The local noise scale is obtained by multiplying the median of the absolute deviations of the data within the local frequency sliding window from the median by a fixed constant factor.

[0040] Since the ODMR signal itself contains concave resonance peaks, the ordinary mean is easily contaminated by extreme points. Therefore, a robust statistical method based on the median and MAD (Median Absolute Deviation) is used to estimate the local dynamic baseline.

[0041] Define tensor elements: The local dynamic baseline mean is defined as: The local noise scale is defined as: in, The MAD estimation effectively suppresses the contamination of mean statistics by broadband impulse noise and is suitable for NV color center ODMR signal processing in high-voltage power grid environments and industrial electromagnetic interference environments. Therefore, the local dynamic baseline is an adaptive robust baseline that dynamically changes with time and frequency.

[0042] S204: For tensor elements Define the adaptive error threshold: in, To set the confidence level, For the corresponding statistical threshold coefficient, This is a local noise scale.

[0043] When the background noise approximately follows a Gaussian distribution: in, It is the inverse cumulative distribution function of the standard Gaussian distribution.

[0044] ODMR extreme value detection is performed using a decision mechanism below the dynamic baseline: in, This indicates that the current frequency point belongs to the candidate resonance peak region; This indicates that the area belongs to the background noise region. For other situations.

[0045] Furthermore, in this embodiment, step S3 performs three-dimensional XOR logic filtering and Hamiltonian eigenvalue trajectory verification on the binarized logic tensor to obtain the denoised target tensor. Specific steps include S301-S304: S301: Along the time dimension, perform an XOR operation on the binary logic tensor for adjacent scan cycles, and determine the elements with an XOR result of 1 as random burst noise and set them to zero. For binary logic tensors Along the time dimension respectively and spatial dimensions Perform mode-unfolding. First, perform an XOR operation on adjacent cycles along the time dimension: If the XOR result is 0 and the logic value is 1, it means that the signal at this frequency point exists continuously in adjacent cycles and is determined as a candidate point for the true resonance peak; if the XOR result is 1, it means that it only appears in a single cycle and is determined as random burst noise and is eliminated.

[0046] S302: Along the spatial dimension, perform Hamiltonian eigenvalue trajectory verification on the retained candidate points, and determine the candidate points whose residuals with the predicted frequency trajectory exceed the preset threshold as deviating from the physical constraints and set them to zero. Since the projection of the same external vector magnetic field onto the four crystal axes has a geometric constraint relationship, if the candidate point of a certain period deviates significantly from the eigenvalue trajectory of the Hamiltonian equation, a second zeroing is performed to obtain the denoised target tensor. .

[0047] Let the actual extraction frequency be... The prediction frequency for the previous period was Then the trajectory residual is defined as follows: If the following conditions are met: This candidate point is considered to deviate significantly from the theoretical trajectory. This is a preset threshold.

[0048] in, ; Estimate the standard deviation for the frequency; The preferred value is 2 to 5.

[0049] If the candidate frequency point does not satisfy the NV space geometric constraints, then perform a second zeroing operation; otherwise, retain it. S303: The Hamiltonian eigenvalue trajectory verification in this embodiment is essentially a spatial consistency verification under the physical constraint of NV color centers and a residual analysis based on the theoretical frequency trajectory and the actual extracted frequency.

[0050] The time continuity prediction is based on the frequency extracted from the previous period and the frequency drift velocity to predict the frequency of the current candidate point. The spin Hamiltonian of the NV color center is defined as: in, For temperature-dependent zero-field splitting parameters, Electron gyromagnetic ratio, This is an external vector magnetic field.

[0051] Under the weak transverse field approximation condition, the first The ODMR frequencies corresponding to each NV crystal axis satisfy: in: For the first Unit vector along the NV crystal axis.

[0052] The spatial geometric consistency constraint is based on the eigenvalue trajectory determined by the projection relationship between the four NV crystal axis directions and the applied vector magnetic field.

[0053] The four NV crystal axes are defined as follows: therefore: Right now: in, Let be the projection component of the magnetic field along the k-th NV crystal axis. This geometric projection relationship forms the basis for spatial dimension correlation constraints.

[0054] S304: To enhance dynamic tracking capabilities, a frequency drift prediction model is further introduced: in, For frequency drift velocity, For scan period; only when The candidate point will only be retained if the time is right.

[0055] Furthermore, in this embodiment, step S4 extracts the resonant frequency values ​​corresponding to each crystal axis from the target tensor and jointly solves the vector magnetic field and temperature using an overdetermined nonlinear equation set. The specific steps include: S401: Using half of the difference between the positive and negative resonant frequencies corresponding to the four crystal axes as the differential modulus, a matrix equation proportional to the projection of the vector magnetic field is constructed, and the components of the applied vector magnetic field are solved by the least squares matrix pseudo-inverse algorithm; at the same time, using half of the sum of the positive and negative resonant frequencies as the common mode, the change in the zero field splitting parameter is separated, and then the temperature change is solved.

[0056] S402: Traversing the denoised target tensor The non-zero elements in the sample are used to extract eight high-precision resonant frequency values ​​corresponding to the four crystal axes in each period, denoted as . .

[0057] The Hamiltonian solving unit constructs an overdetermined nonlinear equation system using the extracted eight frequency values. A least-squares matrix pseudo-inverse algorithm is then used to separate the zero-field splitting parameters caused by temperature variations in real time. The change in magnitude, and the components of the external vector magnetic field in the three directions. The solution results are output via a DAC module or a digital communication interface.

[0058] A linearized calculation method based on the small perturbation approximation is adopted to define the frequency difference mode: but: Further convert to matrix form: in: Therefore, the pseudo-inverse solution of least squares can be used: In the formula, This is the least squares estimate of the magnetic field vector.

[0059] Example 3, referring to Figure 2 This is the third embodiment of the present invention, which differs from the first two embodiments in that: This embodiment also provides a diamond NV color center vector magnetic field measurement system based on spatiotemporal tensor denoising and feature extraction, including: an optical path excitation module, a vector microwave control module, a three-dimensional probe module, and a tensor magnetometry electronics module; The optical path excitation module has an acousto-optic modulator, a first lens, a dichroic mirror, and an objective lens arranged sequentially in front of the laser source; the laser emitted by the laser source is modulated and reflected, and then focused by the objective lens onto the diamond NV color center sample in the three-dimensional probe module. The three-dimensional probe module includes: a three-dimensional orthogonal Helmholtz bias coil, a microwave radiating antenna, a long-pass filter, and a diamond NV color center sample ensemble placed at the center; the three-dimensional orthogonal Helmholtz bias coil is used to apply a known three-dimensional bias static magnetic field to separate the degenerate energy levels of the four crystal axes of the NV color center; The vector microwave control module includes: a multi-channel microwave source, a radio frequency switch, a broadband power amplifier, and a system control PC; the multi-channel microwave source generates a swept-frequency microwave signal covering different crystal axis resonant ranges, which is then fed into the microwave radiating antenna after being amplified by the power amplifier; The tensor magnetometry electronics module includes: a photodetector (PD), a transimpedance amplifier, a programmable bandpass filter, a multi-channel analog-to-digital converter (ADC), an FPGA tensor processing module, and a digital-to-analog converter (DAC). The red fluorescence signal emitted by the diamond NV color center sample passes through the dichroic mirror and long-pass filter, is received by the photodetector and converted into an electrical signal, and is then pre-conditioned by the transimpedance amplifier and the programmable bandpass filter before being input to the multi-channel analog-to-digital converter module. The multi-channel analog-to-digital converter module converts the conditioned continuous fluorescence voltage signal into a discrete digital signal and sends it to the FPGA tensor processing module. The system control PC synchronously sends the three-dimensional bias static magnetic field parameters, microwave sweep period, and frequency point data to the FPGA tensor processing module.

[0060] This embodiment also provides an electronic device, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction proposed in the above embodiment.

[0061] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction as proposed in the above embodiments.

[0062] The storage medium proposed in this embodiment and the diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0063] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0064] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0065] Example 4, refer to Figure 3 This is an embodiment of the present invention, used to verify a method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction.

[0066] In this embodiment, the diamond NV color center vector magnetic field measurement system utilizes a three-dimensional Helmholtz coil to generate a bias static magnetic field of approximately 30 Gauss, causing complete Zeeman splitting of the energy levels of the four crystal axes. When the microwave source scans in the range of 2.7 GHz to 3.0 GHz, modulated fluorescence signals are continuously received at the FPGA via an ADC.

[0067] Compared to traditional one-dimensional arrays or two-dimensional matrices (frequency) The present invention utilizes the FPGA memory mapping mechanism to construct a dimension with different amplitudes. Data tensors .in This represents the number of historical scan cycles in the cache. This refers to the number of frequency points in a single sweep. The projection separation channels correspond to the four crystal axes. Through this tensor quantization, the signal acquires historical memory properties and spatial correlation properties.

[0068] When strong stray electromagnetic interference occurs in the environment, a spike with a width similar to the ODMR resonance peak will be superimposed on the fluorescence signal. If only the two-dimensional matrix method is used to find the minimum value, this spike will be mistakenly identified as the resonance point. In this embodiment, the FPGA along the time dimension... Extracting the tensor tube, i.e. Since the change in the real magnetic field is continuous, the resonance peak occurs over several consecutive operating cycles. and The displacement within will not exceed the allowable error range. .

[0069] Using XOR logic: .

[0070] when When this occurs, it means that the point appears transiently within only one cycle, belonging to non-stationary burst noise. The hardware logic circuit directly sets the tensor slice position to zero. Only when... and Only then is it sent into the spatial dimension. check.

[0071] In spatial verification, based on the Hamiltonian of the diamond NV color center: The system requires that the eight frequencies extracted from the four crystal axes must satisfy the trace constraint (trace invariance). If a set of candidate frequency points does not satisfy the tensor rank constraint, the point with the largest deviation is discarded.

[0072] To demonstrate the significant advantages of the spatiotemporal tensor-based noise reduction algorithm described in this invention over the original data matrix-based two-dimensional analysis method, quantitative experimental simulations and comparisons of the algorithm's application in the relevant field (magnetic field measurement under strong impulse noise environment of high-voltage power grid) were conducted using Python.

[0073] Comparison of experimental results and data analysis: The table below lists a statistical comparison of algorithm performance under the scenario of 50 consecutive dynamic vector magnetic field changes superimposed with strong power grid interference (evaluation metrics include root mean square error RMSE, anti-pulse false peak rate, and vector field solution success rate):

[0074] The experimental data from the above application level show that: Without employing the method of this invention (i.e., using the traditional two-dimensional matrix method), due to the lack of memory judgment of multi-cycle time axes and multi-crystal axis spatial constraints, when instantaneous pulse noise such as grid discharge is mixed in, the system is extremely prone to judging sudden amplitude drops as valid fluorescence signals (leading to a false peak misjudgment rate as high as 67.8%). After extracting the incorrect frequency points and entering them into the Hamiltonian solution module, it will cause the vector matrix solution to fail.

[0075] By using the method of this invention, the data is placed into a three-dimensional tensor. Firstly, random noise is cleanly and efficiently "shielded" by the XOR logic of the time axis (burst pulses cannot continuously reproduce their specific form over multiple cycles). Secondly, the correlation judgment of the spatial axis forcibly corrects the offset. This algorithmic innovation increases the success rate of vector magnetic field calculation from 32.5% to 88.6%, greatly expanding the boundaries of industrial and biomedical applications based on diamond NV color center sensors.

[0076] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction, characterized in that, include: Construct a three-dimensional data tensor, with dimensions corresponding to the time period dimension, the frequency sweep point dimension, and the NV color center crystal axis projection space dimension, respectively. The three-dimensional data tensor is subjected to dynamic adaptive threshold binarization to generate a binary logical tensor; the dynamic adaptive threshold is determined based on the local dynamic baseline mean and the adaptive error threshold under a set confidence level. Perform three-dimensional XOR logic filtering and Hamiltonian eigenvalue trajectory verification on the binarized logic tensor to obtain the denoised target tensor. The resonant frequency values ​​corresponding to each crystal axis are extracted from the target tensor, and the vector magnetic field and temperature are jointly solved by the overdetermined nonlinear equations.

2. The method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction as described in claim 1, characterized in that, The dynamic adaptive threshold binarization process includes: the local dynamic baseline mean and the adaptive error threshold are dynamically estimated by a combination of a local frequency sliding window and a time history buffer window; The local dynamic baseline mean is calculated using a robust statistical method based on the median and absolute median difference, and the adaptive error threshold is generated by multiplying the local noise scale obtained based on the robust statistics with a statistical threshold coefficient determined according to a set confidence level.

3. The method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction as described in claim 2, characterized in that, The frequency local sliding window covers a preset neighboring frequency point with the current frequency point as the center, and the time history cache window includes the current scan cycle and a preset number of previous scan cycles; The mean of the local dynamic baseline is the median of the data within the local frequency sliding window, and the local noise scale is obtained by multiplying the median of the absolute deviation of the data within the local frequency sliding window from the median by a fixed constant factor.

4. The method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction as described in claim 3, characterized in that, The execution of three-dimensional XOR logic filtering and Hamiltonian eigenvalue trajectory verification includes: performing XOR logic operation on the binarized logic tensor along the time dimension for adjacent scan cycles, and determining the element with an XOR result of 1 as random burst noise and setting it to zero; Along the spatial dimension, Hamiltonian eigenvalue trajectories are verified for the retained candidate points. Candidate points whose residuals with the predicted frequency trajectory exceed a preset threshold are judged as deviating from physical constraints and set to zero.

5. The method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction as described in claim 4, characterized in that, The Hamiltonian eigenvalue trajectory verification includes: temporal continuity prediction and spatial geometric consistency constraint; wherein, the temporal continuity prediction is based on the frequency extracted in the previous period and the frequency drift velocity to predict the frequency of the current candidate point, and the spatial geometric consistency constraint is based on the eigenvalue trajectory determined by the projection relationship between the four NV crystal axis directions and the applied vector magnetic field.

6. The method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction as described in claim 5, characterized in that, The method of jointly solving the vector magnetic field and temperature by using an overdetermined nonlinear equation system includes: using half of the difference between the positive and negative resonant frequencies corresponding to the four crystal axes as the differential modulus, constructing a matrix equation proportional to the projection of the vector magnetic field, and solving for the external vector magnetic field components by using the least squares matrix pseudo-inverse algorithm; at the same time, using half of the sum of the positive and negative resonant frequencies as the common mode, separating the change in the zero-field splitting parameter, and then solving for the temperature change.

7. The method for measuring the vector magnetic field of diamond NV color centers based on spatiotemporal tensor denoising and feature extraction as described in claim 6, characterized in that, The least squares matrix pseudo-inverse algorithm is solved based on a geometric matrix composed of unit vectors of the four NV crystal axis directions; the temperature change is obtained by dividing the change of the zero-field splitting parameter by the temperature coefficient.

8. A diamond NV color center vector magnetic field measurement system based on spatiotemporal tensor denoising and feature extraction, employing the method for measuring the diamond NV color center vector magnetic field based on spatiotemporal tensor denoising and feature extraction as described in any one of claims 1 to 7, characterized in that, include: Optical path excitation module, vector microwave control module, three-dimensional probe module, and tensor-quantized magnetometry electronics module; The optical path excitation module has an acousto-optic modulator, a first lens, a dichroic mirror, and an objective lens arranged sequentially in front of the laser source; the laser emitted by the laser source is modulated and reflected, and then focused by the objective lens onto the diamond NV color center sample in the three-dimensional probe module. The three-dimensional probe module includes: a three-dimensional orthogonal Helmholtz bias coil, a microwave radiating antenna, a long-pass filter, and a diamond NV color center sample ensemble placed at the center; the three-dimensional orthogonal Helmholtz bias coil is used to apply a known three-dimensional bias static magnetic field to separate the degenerate energy levels of the four crystal axes of the NV color center; The vector microwave control module includes: a multi-channel microwave source, a radio frequency switch, a broadband power amplifier, and a system control PC; the multi-channel microwave source generates a swept-frequency microwave signal covering different crystal axis resonant ranges, which is then fed into the microwave radiating antenna after being amplified by the power amplifier; The tensor magnetometry electronics module includes: a photodetector (PD), a transimpedance amplifier, a programmable bandpass filter, a multi-channel analog-to-digital converter (ADC), an FPGA tensor processing module, and a digital-to-analog converter (DAC). The red fluorescence signal emitted by the diamond NV color center sample passes through the dichroic mirror and long-pass filter, is received by the photodetector and converted into an electrical signal, and is then pre-conditioned by the transimpedance amplifier and the programmable bandpass filter before being input to the multi-channel analog-to-digital converter module. The multi-channel analog-to-digital converter module converts the conditioned continuous fluorescence voltage signal into a discrete digital signal and sends it to the FPGA tensor processing module. The system control PC synchronously sends the three-dimensional bias static magnetic field parameters, microwave sweep period, and frequency point data to the FPGA tensor processing module.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the diamond NV color center vector magnetic field measurement method based on spatiotemporal tensor denoising and feature extraction as described in any one of claims 1 to 7.

Images